As agents have taken over the AI scene in 2024, companies like LlamaIndex that create developer tools for orchestrating complex agentic LLM applications have learned valuable lessons about the best way to implement agentic behavior.
In this worshop, we’ll cover the latest and greatest implementation used by LlamaIndex Workflows: Event-Driven Architecture (EDA). This technique for creating agentic applications stands out above the chain and pipeline implementations that use Directed Acyclic Graphs (DAGs) as well as the cyclic graph approaches that have been gaining in popularity.
During this workshop, we’ll break down each of the core constructs within LlamaIndex Workflows, which fundamentally are made up of steps (i.e., ****Python functions). Each step is responsible for handling certain event types and emitting new events.
Join us to get out ahead of the hackathon and learn the the concepts and boilerplate code you need to understand and build a complex agentic application with LlamaIndex!
🤗 Hugging Face
You will need access to a Hugging Face account to download model weights
💬 OpenAI API
You must enable billing through the OpenAI API, which we will use throughout the workshop.
🌲 Pinecone
🦙 (Optional) Llama 3
We’ll use completely open-sourced versions of Llama 3.1 (e.g., from NousResearch). However, to use the models directly from Meta, you’ll need to share your contact information with Meta. You can find the Meta model here.
We’ll cover: